Geographic Variation in Leach's Storm-Petrel

Home , List of individual birds, Northern fulmar, Petrel

GEOGRAPHIC VARIATION IN LEACH'S STORM-PETREL

DAVID G. AINLEY Point Reyes Bird Observatory, 4990 State Route 1, Stinson Beach, California 94970 USA

ABSTRACT.-•A total of 678 specimens of Leach's Storm-Petrels (Oceanodroma leucorhoa) from known nestinglocalities was examined, and 514 were measured. Rump color, classifiedon a scale of 1-11 by comparison with a seriesof reference specimens,varied geographicallybut was found to be a poor character on which to base taxonomic definitions. Significant differences in five size charactersindicated that the presentlyaccepted, but rather confusing,taxonomy should be altered: (1) O. l. beali, O. l. willetti, and O. l. chapmani should be merged into O. l. leucorhoa;(2) O. l. socorroensisshould refer only to the summer breeding population on Guadalupe Island; and (3) the winter breeding Guadalupe population should be recognizedas a "new" subspecies,based on physiological,morphological, and vocal characters,with the proposedname O. l. cheimomnestes. The clinal and continuoussize variation in this speciesis related to oceanographicclimate, length of migration, mobility during the nesting season,and distancesbetween nesting islands. Why oceanitidsfrequenting nearshore waters during nestingare darker rumped than thoseoffshore remains an unanswered question, as does the more basic question of why rump color varies geographicallyin this species.Received 15 November 1979, accepted5 July 1980.

DURING field studies of Leach's Storm-Petrels (Oceanodroma leucorhoa) nesting on Southeast Farallon Island, California in 1972 and 1973 (see Ainley et al. 1974, Ainley et al. 1976), several birds were captured that had dark or almost completely dark rumps. The subspecificidentity of the Farallon population, which at that time was O. l. beali, was thus problematic, because the presence of dark-rumped birds indicated inclusion in O. l. willetti (A.O.U. 1957). Not long after the capture of these birds, the A.O.U. (1973) synonymized O. l. willetti with O. l. beali. This action, following the recommendation of Austin (1952), "solved" the immediate question, but the problem seemedworthy of further pursuit, as Crossin (1974) found it impossibleto assignto any of five subspeciesmost specimenscollected away from breeding sites. He thought that most subspeciesof O. leucorhoa should be synonymized, a recommendation already made by Loomis in 1918. Van Rossem (1942), however, questionedthe wisdom of synonymizingall subspecies,and, exceptfor the recent decisionon O. l. willetti, the A.O.U. (1957, 1973) has followed van Rossem's subspecificdefinitions. The present study tests Loomis' (1918) and Crossin's(1974) hypothesis that Leach's Storm-Petrel in the eastern North Pacific is a dichromatic speciesin which the proportions of color phasesand mensural charactersvary geographically in a cline, rather than there being the series of distinct populations currently recognized.I also compared vocalizationsamong birds from several localities and tried to reveal those factors that might account for geographic variation in this species.

METHODS

Eight measurementsplus a classificationof rump color were made on 514 specimensor live individuals separatedinto 13 samples.The Farallon Island sample included 48 live birds mist-netted on the nights of 29-31 May 1978. All other samplesconsisted only of study skins. The rump color of an additional 164 skins from Guadalupe Island was classifiedin order to clarify the relationshipsof the summer and winter breeding populations there. Only specimenscollected on or immediately adjacent to a known breeding locality (i.e. the birds that flew aboard a boat anchored within a few hundred meters of shore) were used for measurement. Besides six standard measurements(Table 2), the shape (roundedness)of the wing tip was assessedby measuringthe distancefrom the tip of primary 10 to the tips of 9 and 7; the greater the

837 The Auk 97: 837-853. October 1980 838 DAVIDG. AINLEY [Auk, Vol. 97

Fig. 1. The seriesof referencespecimens (left to right, 1 to 11) showingcolor variation in O. leu- cotboa. differencein measurementthe more pointed the wing. Statistical significance between samples was tested usingDuncan's new multiple range test for a one-wayclassification of variance (Steel and Torrie 1960). Sampleswere compared at the 1% significancelevel. I found,as did Crossin(1974), that femaleLeach's Storm-Petrels were slightly larger than males but thatthe difference was not statistically significant without very large sample sizes (over 80 individuals), andit did notvary geographically. I thus combined the two sexesfor analysis;within all samplesthe sexeswere represented about equally. This combinationallowed me to useunsexed specimens, helped to increasesample sizes, and removed the problemof usingincorrectly sexed specimens. Rumpcolor was classed by comparingspecimens with a seriesof 11reference skins. Each represented an approximatelyequal step in a continuumbeginning with all uppertail covertsentirely white, except for thefeather shafts, which may or maynot have been white (class 1), andending with all uppertail covertsentirely dark or concolorouswith the back and tail (class11). Specimens in the series (Fig. 1), all fromthe NationalMuseum of NaturalHistory, were as follows: 1 = USNM 201458,2 = 543662,3 = 132761,4 = 545030,5 = 544507,6 = 545031,7 = 543697,8 = 543694,9 = 543708,10 = 543726,11 = 543659.As for the intensityof a "bluishbloom" on the brownbody color, a characterthat wasgiven greatimportance in originaldescriptions of varioussubspecies and wasmost recently alluded to by Bourne(in Palmer 1962),I agreewholly with Crossin(1974) that it relatesto the extentof featherwear andthe ageof specimens(not the ageof thelive bird)rather than to geographicvariation. Recordingsof Leach'sStorm-Petrel vocalizations were made by R. L. DeLong,Pacific Ocean Biological SurveyProgram (POBSP), at theCoronado and San Benito islands in May 1967and at GuadalupeIsland in Juneand October1967, using a Uher Report-L4000 tape recorderand microphone.I recorded vocalizationsat the FarallonIslands in May 1975,using a SonyTC-800B recorder and an Altecmicro- phone.Sonagrams of Flight Calls and burrow Chatters were made at theMuseum of Vertebrate Zoology, Berkeley,using the wideaudio band on a Kay InstrumentCompany Sonagraph. Populationsat GuadalupeIsland.--Two populationsof O. leucorhoabreed at different seasonson GuadalupeIsland, but informationabout them, summarized by Crossin(1974), is fragmentary.I con- sideredspecimens collected on or nearthe islandfrom Octoberthrough April to be from the winter populationand thosecollected from May throughSeptember to be from the summerpopulation. This wasdeduced by comparingthe proportions of colormorphs from month to month(Table 1), a procedure suggestedby Crossin'sdiscovery that theseproportions seemed to differbetween the two populations. A clearseparation became obvious (Fig. 2):the colorsof summerbirds were spread between classes 1- 11but concentrated bimodally between 2-3 and7-10. Supportingthis separation were the facts that (1) theonly September specimens in the 10collections searched were recently fledged (collection dates were withinthe last few daysof that month);and (2) observationsof fledglings by otherpersons, summarized by Crossin,also occurred late in Septemberwith occasional dates as late as early November. It appears thatthe peak of fledging of summer chicks occurs in verylate September and early October. Extrapolating October1980] Leach'sStorm-Petrel Taxonomy 839

TABLE 1. The number of individuals of each color morph in samples of O. leucorhoa collected at Guadalupe Island.

Colormorphs a Total speci- Months 1 2 3 4 5 6 7 8 9 10 11 mens

October 3 6 12 10 6 1 1 39 November 1 3 5 7 1 17 December 1 4 4 10 8 3 31 January 2 2 4 4 4 1 17 February 1 1 3 2 3 2 1 13 March 1 (1) (1) 2 2 1 1 9 April 1 (1) 1 1 4 May 1 8 1 1 11 June 2 12 12 2 2 2 7 11 15 7 2 74 July 7 1 2 1 2 1 14 August 1 5 1 6 9 9 3 1 35 September (1) (1) (1) 3 Total 267

a Numbers in parenthesesare fledglings.

from informationon breedingphenology at the Farallons(Ainley et al. 1974), egglaying of the summer breeding population of Guadalupe should occur from late May to the end of June. Crossin (1974: 174) surmisedthe egg-layingpeak to be late in the third week of June 1968, when the POBSP made a visit to Guadalupe. As for the winter population, information given by Crossin (1974: 174), plus the dates of three fledging specimensI inspected,suggested that most chicksfledge from early March to mid-April. The latest April specimenwas recently fledged and the other three April specimens(adults) were at a stage of molt or showed a degree of feather wear that, judging from information on molt in O. leucorhoa at the Farallons(Ainley et al. 1976),clearly placedthem in the winter breedingpopulation. Again using the Farallon data, egglaying in the summerpopulation of Guadalupeshould occur from mid-November through December. Crossin reported no eggspresent during a POBSP survey on 21 October 1967. The secondtype of information that supported my division of breeding populations is the fact that no adult specimensof O. leucorhoa have been collected(or at least were among the 267 specimensI examined) at Guadalupe during the periods 25 April-28 May and 31 August through mid-October. This is consistentwith the suppositionthat May and October mark the initial month of the summerand winter breedingseasons, respectively. During thosemonths adults would make sporadicvisits at night to court and prepare nest sites and would not remain during the day. The few adults from the previous breeding seasonsstill providing for late chicks would visit only for a few hours at night, and probably not every night, to feed their chicks (which would be near to fledging). Thus, personsintent on collectingstorm- petrels, and who would probably undertake the hazardous trip ashore only during daylight, would be unlikely to find petrels. Hence, fewer specimensare available from the initial months of breeding seasons. On the other hand, I do not know how many collectorsvisited Guadalupe during those periods;the POBSP, however, found only six storm-petrelsduring the day on their October visit.

RESULTS

Rump color variation.--The proportions of rump color classesgradually changed with decreasinglatitude, the Guadalupe samplesexcepted (Fig. 2). From the Aleu- tians to Sitka (St. Lazaria Is.), as well as in the North Atlantic, classesranged from 1 to 4, and 2 and 3 were dominant. From Forrester Island, Alaska south to northern California, color classesranged from 1 to 5; initially 2 and 3 prevailed, but with decreasinglatitude 4 and 5 increasedand class2 decreased.This shift toward darker rumps continued, with much larger steps, from the Farallons to the San Benitos. Only two skins from San Miguel Island were available, but four birds mist-netted and released in 1976-77 were "white-rumped" (S. Speich pers. comm.), i.e. among classes2-7. As for Guadalupe Island, the distribution of classesdiffered between the two samples (winter vs. summer, see Methods) and showed no relationship to 840 DAVID G. AINLEY [Auk, Vol. 97

47 ATLANTIC• T--T'•• ' I NO. CALIFORNIA

52 ALEUTIANIS. •5I I , • •• FA[RALLONIS.

ST. LAZARIA IS, • •-•

z z

COLOR CLASSES

Fig. 2. The proportion of color classesin samples of O. leucorhoa. the clinal change evident in the other Pacific samples. The Guadalupe winter and Farallon samples were very similar except that only the latter contained the very darkest rump classes.The pattern in the Guadalupe summer sample resembled none of the others. Except for Guadalupe, where samplesobviously differ from one another and from all others, Loomis' (1918) proposal appears more or less correct, i.e. that Leach's Storm-Petrels in the eastern North Pacific are a single polychromatic population in which the proportion of dark phase (i.e. dark-rumped) individuals increases with decreasinglatitude. Thus, although rump color was central in the original descrip- tions of several subspeciesof Leach's Storm-Petrel, it should be accorded little more taxonomic significancethan in other procellariiformesshowing similar color variation, for example the Northern Fulmar (Fulmarus glacialis, Palmer 1962: 142-143), Wedge-tailed Shearwater (Puffinus pacificus, Murphy 1951), and White-throated Storm-Petrel (Nesofregetta albigularis, Crossin 1974). This is especially so because every possiblegradation between totally white- and totally dark-rumped individuals occurs. Furthermore, excluding Guadalupe samples, the continual shift in proportions of color variants indicates that some factor selecting for dark rumps comes gradually to bear on populations, or converselythe factor selecting for white rumps lessens,rather than there being distinct, all-inclusive differencesfrom one population to another. The fact that Guadalupe populations depart markedly from the pattern shown in adjacent coastal populations may help to explain the color variation. Size variation.--Average measurementsof all characters were largest in the Aleu- tian Island and Atlantic samples (Table 2). In the latter, the tail fork was deeper than in many but not all Pacific samples. Within the coastal Pacific samples(Gua- dalupe not included), measurementschanged gradually with decreasinglatitude, but first a decrease occurred through several samples, followed by an increase. For example, the culmen was largest in the Aleutian (52ø-59øN) and St. Lazaria (57ø) samples, decreasedto Forrester Island (55ø), remained at about that level to northern October 1980] Leach'sStorm-Petrel Taxonomy 841

California, and then increased to a level shared by the Farallon to San Benito samples. The same general pattern was apparent in all other characters, except tarsus length and wing shape. Tarsus length gradually and consistentlydeclined with decreasinglatitude, with the exceptionof the Coronados'sample, where it was inconsistentlylarge. Wing shape showed little in the way of gradual change; rather, samplesseparated into two groups, birds with pointed wing tips (Atlantic and Aleu- tians to northern California) and birds with round wing tips (Farallon to San Benito). Within the latter samples,roundedness increased with decreasinglatitude. Size characters in the Guadalupe sampleswere consistentlysmaller than in almost all others. The summer breeding birds were, in turn, always smaller than the winter birds. The summer birds had the roundest wings of all samples;wing shape in winter birds was similar to the Farallon sample. Significant differencesin the five most reliable mensural characters are summarized in Table 3; due to large CV's, depth of tail fork and wing shapeare excluded. From the Aleutians to northern California, and to a lesserdegree to the San Benitos, the decreasingnumber of significant comparisonsas one moves from left to right in the table indicates that samples are more easily separated as the actual geographic distance between sample sites increases.In most casesbirds from adjacent colonies or nearby regions cannot be separated. There are, in addition, several unsampled nesting sites between the Aleutians and St. Lazaria (Sitka) and at least one between the Farallons and Coronados(San Miguel Island). Quinlan (pers. comm.) recently measured birds from the Prince William Sound area and found measurements to be intermediate between Aleutian and Sitka samples. This demonstrates further the clinal nature of geographicvariation in this species.Is it possiblethen to distinguish any of the samples from the others? Samples from St. Lazaria, Coronados, San Benitos, and Guadalupe (both seasons) are distinguishableat the 1% level from at least one other sample by at least one of five "reliable" characters (Table 3). It would be folly, however, to propose six subspecies, because placing an individual into one of the four presently recognized is virtually impossible(Austin 1952, Crossin 1974)! Here is a good example of why taxonomy at or below the specieslevel in clinally varying populations is not the same problem in seabirds as it is in most landbirds. Seabirds are highly philopatric, and, because their breeding localities are islands (which, naturally, are disjunct), geneflow is very much disrupted and fragmented. When environmental factors vary slightly from one island to another, slight morphologicaldifferences between samples result. Often, as in the present species,these differencesare recognizableonly when the breeding locality of a specimenis known. The problem in Leach's Storm-Petrels is aggravated by the extensive range in latitude and environment of nesting islands (see Discussion). Such a range is shown in few other seabirds (another example is Puffinus puffinus). The resulting slight average differencesin morphology between samples, which led to the descriptionof an array of subspecies,no doubt underlies the taxonomic difficulties I am trying to overcome in this study. Within Table 3, three groupingsof samplesemerge. The first, and largest, includes all samplesexcept thosefrom Guadalupe Island. Each sample is either similar in all respectsto at least one other sample (e.g. Aleutian and eastern Atlantic samples)or is distinguishablefrom other samplesonly by virtue of one character(e.g. San Benito sample). In casessimilar to the latter, while one character might on the average be different from another sample in the group, never doesit differ consistentlyfrom all 842 DAVIDG. AINLEY [Auk,Vol. 97 October1980] Leach'sStorm~Petrel Taxonomy 843

+t7q +•7• +• +•7• +•7' +•7• +17•

+•7• +I7m+• +•7• *17q+• 844 DAVIDG. AINLEY [Auk,Vol. 97

TABLE 3. Number of reliable mensuralcharacters (bill depth and culmen, wing, tarsus, and tail length) distinct (P < 0.01) when each sample is compared against all others using Duncan's new multiple range test.

North Atlantic - Aleutian Islands 0 a -- St. Lazaria Island 4 1 - Forrester Island 5 0 1 - British Columbia 5 3 2 0 - Washington 5 4 4 1 0 - Oregon 5 2 1 0 0 0 Northern California 5 4 2 1 0 0 1 - Farallon Islands 5 2 1 1 2 3 1 2 - Coronado Islands 5 1 2 1 1 1 1 2 0 - San Benito Islands 5 3 1 1 1 1 2 1 1 1 - Guadalupe Island, Summer 5 5 5 5 5 5 5 5 5 5 4 Guadalupe Island, Winter 5 5 4 4 4 3 4 4 5 4 4

a Significantat P < 0.05. samples. The Guadalupe samples, however, are easily distinguishable. They differ from each other and from all other samples on the basis of at least three, and often more, characters. Van Rossem(1942), Austin (1952), and Crossin (1974) all agreed that Guadalupe O. leucorhoa are distinct, but as first surmised by Crossin and shown here, the winter and summer populations are far more different morphologically from one another than are samplesfrom the Aleutians to San Benitos. Variation in vocalizations.--No studies have critically analyzed the behavioral significanceof vocalizations in storm-petrels. In my experience with five species, two important calls, and the two that are perhaps the most frequently used, are Chattering and the Flight Call. The first is given only in the burrow, usually when both members of the pair are present, and often they duet. The message(cf. Smith 1977) in the call must have to do with association. The second, the Flight Call, is usually given as a bird flies about in the colony, especially when it passesover its burrow in a repeating circle. It is also given from the burrow, especially when a flying bird calls nearby. These and other contextssuggest this call to be the storm- petrel song, in the classicalsense of territory and self-declaration. Tape recordingsof Chattering in O. leucorhoawere available from five populations. Some qualitites of the calls are compared in Table 4, and sonagramsare presentedin Fig. 3; sonagramsof calls from North Atlantic O. leucorhoa are pre- sented in Hall-Craggs and Sellar (1976). When Chattering, a storm-petrel utters a rapid series of chirps or clicks, ranging from about 0 to 2 kHz, and then produces a long wheeze as it takes a breath. Series of chatters and wheezes may go on continuously for many minutes. In the Atlantic and coastal Pacific samples, the frequencyof chattersis 15-23/s (2 = 19.6), and a breath lasting about 0.4 s is taken every 2-4 s. When the bird is taking a breath, the wheeze is as loud as the chatters. All chatter notes are approximately the same sound except that the first one after a breath is a bit stronger than the others. In the two Guadalupe populations, the October 1980] Leach's Storm-Petrel Taxonomy 845

TABLE 4. Characteristicsof burrow Chattering in different Leach's Storm-Petrelpopulations.

Time between breath Breath Breath Note at Calls/ Chat- onsets duration note start vs. Population individual ters/sa (s) (s) audibleb others Iceland½ 4+/4+ 15-22 3.1-4.8 0.4-0.6 Yes Slight emphasis, same quality Farallon Islands 12/5 17-21 1.0-3.0 0.2-0.4 Yes Slight emphasis, (18) same quality San Benitos Islands 7/3 22-23 3.3-3.9 0.40 Yes Slight emphasis, (22) same quality Guadalupe, summer 4/2 26-27 5.1 0.6 Barely Strongemphasis, (26) different note Guadalupe,winter 9/4 25-27 4.3-4.6 0.6-0.8 Barely Strongemphasis, (26) different note

The rangeand mean numberof chatters/sare given. To human ears. From Hall-Craggs and Sellar 1976.

frequency of chatters is more rapid, ranging 25-27/s (• = 26.0), and a breath is taken every 4-5 s. The wheeze at inhalation is barely audible and lasts about 0.6 s. The first note after inhalation is distinctly different from the chatters that follow and is much louder. Chattering in the two Guadalupe populations is similar, and it differs from that in the other samples. Tape recordings of the Flight Call in O. leucorhoa were analyzed for six populations (Table 5 and Fig. 4). In all casesthe length of the call was about equal and notes ranged from 0.5 to about 3.5 kHz. The quality of notes was also similar, except for the Guadalupe winter birds, whose notes were raspy and much less "melodious." Other major differences arose in the total number of notes in a call and the location of the accent or emphasis. In North Atlantic and Pacific coastal populations, calls consistedof 9-12 notes(usually 11) with accentsplaced on 3 notes, 1 at the start and 2 in the middle. The call of Guadalupe summer birds was much different. There were many more notes (11-20 but usually 17), and emphasis was placed on only 1 note about two-thirds through the sequence.In the winter birds, not only were notes more numerous and very raspy, but emphasis was placed on 4 notes, 1 at the start and 3 at the end. Systematic revision.--The results indicate that the taxonomy of O. leucorhoa should be reorganized. The revision proposedbelow lessensthe problems of assign- ing specimenscollected or individuals encounteredat seato appropriate populations, a difficulty inherent in the presently accepted scheme(Austin 1952, Crossin 1974). The proposedrevision emphasizesthe evolutionary pressuresthat account for geographic variation in these populations, an important goal in modern systematics (Selander 1971).

Oceanodroma leucorhoa leucorhoa Viellot: ocean near Picardy, France. This population breeds in the North Atlantic and in the North Pacific from the Aleutian Islands northeast to southeast Alaska and south on coastal islands to Islas San Benitos, Mexico (and probably southwestfrom the Aleutians to southernJapan 846 DAVID G. AINLEY [Auk, Vol. 97

5G. c

i.t.i

Fig. 3. Sonagrams of Chattering in O. leucorhoa nesting at the Farallon Islands (A); San Benito Islands (B); and Guadalupe Island, summer (C) and winter (D).

TABLE 5. Characteristics of Flight Calls in different Leach's Storm-Petrel populations.

Number Calls/ Duration Total emphasis Emphasis Population individual (s) notesa notes notes Iceland b 1/1 1.3 10 3 1, 4, and 5 Farallon Islands 9/3 1.2-1.3 10-12 3 1,4, and5 or (10.7) 1, 5, and 6 Coronados Islands 2/1 1.3 12 3 1, 5, and 6 San Benitos Islands 16/7 1.2-1.5 9-12 2-3 1, 4, and5 or (10.7) 1, 5, and 6 or 1 and 6 Guadalupe Island, summer 8/6 1.2-1.6 11-20 7 (16.8) 16 11 15 12 13 Guadalupe Island, winter 4/2 1.4 14 4 1 12 14

The range and mean number of notesare given. From Hall-Craggsand Sellar 1976. October1980] Leach'sStorm-Petrel Taxonomy 847

Fig. 4. Sonagrams of Flight Calls in O. leucorhoa nesting at the Farallon Islands (A); Coronado Islands(B); San Benito Islands(C); and GuadalupeIsland, summer(D) and winter (E). and Korea). Individuals nesting on North Atlantic islands winter and molt in the equatorial Atlantic. Individuals nesting on Pacific islands do likewise in the equatorial Pacific, with dark-rumped birds wintering at the eastern portion of this region (Crossin 1974, Ainley pers. obs.). Size varies continuouslyand clinally from the Atlantic and Aleutian islands south to the San Benitos. The largest individuals, those from the Aleutians and North Atlantic, have white rumps. South of the Aleutians, rump coloration darkens at first 848 DAVIDG. AINLEY [Auk, Vol. 97

very gradually but acceleratesrapidly with decreasing latitude beginning at the Farallons;most birds at the southernextreme of the population'srange have entirely dark rumps. Much the same trend apparently occurswith decreasinglatitude in the western Pacific (Austin 1952), but further study is required. The Flight Call, or song, is virtually the same throughout the range in the Atlantic and the northern and eastern Pacific, as is the Chattering vocalization. This grouping thus synonymizesO. I. beali Emerson (type locality Sitka, Alaska) and O. I. chapmani Berlepsch (type locality San Benito Is., Baja California) under O. I. leucorhoa and supports the recent incorporation of O. I. willetti van Rossem (type locality CoronadosIs., Baja California)into O. I. beali (0. I. leucorhoa).

Oceanodroma leucorhoa socorroensis Townsend: ocean near Socorro Is., Mexico (Synonym: O. I. kaedingi Anthony). This population breedson isletsoff Guadalupe Island, Mexico, and perhapsfor- merly on the main island itself, during the summer (from May into October). Birds winter and molt in the equatorial Pacific, primarily in the eastern portion of that region (Crossin 1974). Individuals are smaller in all measurementsthan in any other population. Rump color indicates all variations with two distinct peaks, one at the white and the other at the dark end of the scale. The song of O. l. socorroensisis similar to but clearly distinct from that of all other subspecies.Another call, Chat- tering, is similar in this and the following subspeciesbut is different from that in the others.

Oceanodroma leucorhoa cheimomnestes subsp. nov. Type: Adult male; USNM 305762, among the earliest collected specimensI inspected;Guadalupe Island, Mexico; 2 March 1911; collectedby P. I. Osburn. Subspecific characters: Similar in color variation to O. I. leucorhoa in central California (Farallon Islands), except that individuals have at least some white in feathers at either side of the rump, i.e. totally dark individuals are lacking. Most measurements smaller than in all other populations except O. I. socorroensis;in particular, wing and tarsus length separate individuals from northern and bill mea- surementsfrom southern O. I. leucorhoa. Differs morphologicallyfrom O. I. socorroensis by having a distinctly longer bill, tail, and tarsus and more pointed wings. The songis distinct from that in all other populations, and the Chatter vocalization is similar only to O. l. socorroensis.The only temperate breeding storm-petrel known to nest during the winter. Measurements of type: wing 142; culmen 14.8; bill at base 5.4; tarsus 22.3; tail 71.0; depth of tail fork 15.2; color class 4. Specimensexamined: AMNH 131326, 131324, and USNM 305763, collected 2 March 1911, plus many specimens,all collectedmore recently, in USNM, AMNH, MVZ, LACM, SDNHM, and one at CAS (28152); AMNH 131317, a male, collected in the "vicinity of Guadalupe Island" on 5 March 1911 is identifiable as O. I. leucorhoa. Remarks: Crossin (1974), who suspectedthe distinctiveness of this population, proposedthat the name O. I. kaedingi Anthony be reinstated for it. This name had earlier been synonymizedunder O. I. socorroensis(van Rossem1942, A.O.U. 1957). The type of kaedingi (Carnegie Museum 22219) and a large series of paratypes, all October1980] Leach'sStorm-Petrel Taxonomy 849

collected at sea north of Guadalupe on 25 July 1897, are from the summer population, as is the type of socorroensis(USNM 117497; wing 140, culmen 14.0, bill at base5.0, tarsus19.6, tail 70.0, depth of fork 12.5). Becausesocorroensis was described before kaedingi, its name takes priority for Guadalupe'ssummer population, leaving the winter population unnamed. Three other oceanic birds also exhibit temporally distinct breeding populations: Pterodroma mollis in Madeira (Bourne 1957), Macronectes (giganteusand halli) on Macquarie Island and other subantarctic islands (Bourne and Warham 1966), and O. castro in the Galapagos (Harris 1969). The last shows no morphological or behavioral differencesbetween populations, and thus there is no reason to consider separate taxonomic status. The two populations of P. mollis are distinct in several regards, but, as they never meet, Bourne (1957) maintained their subspecificstatus. The two temporal populations of Macronectes, however, nest side by side without interbreeding on Macquarie, their breeding being a few weeks out of synchrony, and they exhibit minor morphologicaland behavioral differencesas well. Bourne and Warham (1966) recommended that they be recognized as distinct but sibling species. On Guadalupe, the two populations of O. leucorhoa are morphologically and behaviorally distinct, but, as they do not meet one another, their situation is somewhat similar to that of P. mollis. For consistencythen, separate subspecific status is appropriate. On the other hand, for the sake of argument, so different are their songsthat, if they met, it is questionablethat interbreeding would occur. Much more work is needed on vocalizations in O. leucorhoa. Etymology: from the Greek, cheimon, winter, and ranestes, suitor, a masculine noun; cheimomnestesrefers to the winter breeding seasonof the population.

DISCUSSION

Factors known to affect geographicvariation in landbirds were reviewed by Se- lander (1971), and I attempt here to relate some of them to Leach's Storm-Petrels. Tarsus and wing length have been used as measuresof body size, and, using such information with air temperatures, tests of Bergmann's ecogeographicrule have been made. A comparisonof tarsus and wing length of storm-petrelsamples to ocean temperaturein the vicinity of nestingareas (Table 2) indicatesagreement with Berg- mann's rule in the eastern North Pacific. Given the rather general nature of the data for ocean temperature, (i.e. a 1-month average for 1971 taken from a large-scale isothermmap) and the fact that at least one other factor stronglyaffects wing length (through wing shape, seebelow), it is understandablethat correlationbetween wing length and temperature is not quite significant(r = 0.5765, P > 0.05). If two rather divergent samples, Coronadosand San Benitos, are disregardedbecause their very rounded wing tips (see below) probably have affected wing length, then a very significant negative correlation exists (r = -0.7901, P < 0.01). At Guadalupe, where the two breeding populationsexperience different temperatures, their respec- tive tarsus and wing lengths are in agreement with the expected responseto temperature. Why O. leucorhoa should be slightly larger in the western North Atlantic than in the northern North Pacific is not obvious from a comparisonof the available sea- surface temperature data. Average August temperaturesare the same for both regions (Sverdrup et al. 1942: chart IV). On the other hand, temperatures are much colder in the western North Atlantic area earlier in the year (Sverdrup et al. 1942: 850 DAVIDG. AINLI•¾ [Auk, Vol. 97

NO.BIRDS

130 ø I10 o

Fig. 5. The localities of dark storm-petrels(all species)seen or collected(summarized from reports in Crossin 1974).

chart III). It is likely that storm-petrels in the Atlantic begin nesting under colder conditionsthan they do in the Aleutians. Climate is also generally more severe in the Atlantic than the Pacific (see Briggs 1974) and, on these grounds, the larger size of North Atlantic O. leucorhoais explainable. Another morphologicalcharacteristic that varied geographicallywas wing shape. It has been demonstrated that some terrestrial long distance migrants have longer or more pointed wings than the shorter distance migrants (seereview in Dorst 1962). This could also be so in Leach's Storm-Petrels, although the trend may be confound- ed somewhat by differencesin foraging distancesduring breeding, which for some populations could involve hundreds of kilometers (see below). Terrestrial migrants rarely forage so widely! The major wintering area for Pacific O. leucorhoa is the equatorial Pacific (Crossin 1974, Ainley pers. obs.). More northerly breeding individuals would have longer distancesto travel between wintering and summering areas, and having longer, more pointed wings would assistthem during migration. If and when rump color variation can be explained, the following points will probably have to be considered.(1) Three other entirely dark oceanitidsnest sympatrically with thoseof Leach's"populations" that include dark individuals but not with those that have only light individuals; these speciesare the Black Storm-Petrel (0. melania), Ashy Storm-Petrel (O. homochroa)and Least Storm-Petrel (Halocyp- tena microsoma).(2) All dark-rumped forms, including dark-rumped Leach's, occur only near the coast during summer (Fig. 5). (3) Dark oceanitids winter nearer the October1980] Leach'sStorm-Petrel Taxonomy 851

coast and white-rumped Oceanodroma winter in the central Pacific. (4) Dark- rumped individuals breed at Guadalupe during summer (and not during winter) when other dark oceanitids are also breeding. (5) The sampleshaving a large per- centage of dark forms have rounder wing tips than the birds from predominantly white-rumped samples. The first three of these five "coincidences"were also noted by Crossin(1974). Somefactor(s), then, is (are):(1) selectingfor dark-rump coloration in storm-petrelsnesting during the summer from the Farallons to the San Benitos, including Guadalupe Island; (2) selectingmore strongly for this factor with decreasing latitude; (3) selectingless strongly for it during winter; and (4) selectingfor it in populations that are less mobile or short-ranging. Consistent with these is the fact that the white-rumped winter Guadalupe population has more pointed wings than the more dark-rumped summer one. It is also interestingthat another white-rumped storm-petrel, O. macrodactyla, until recently bred at Guadalupe during the summer. If like other white-rumped forms it fed at great distancesfrom the nesting site, its presenceoffshore may have made it more opportune for some O. leucorhoa to feed close by. This would be added selection pressure for lessenedmobility in the latter. Oceanodromamacro- dactyla may have had another effect on O. leucorhoaas well: its large size probably favored the small size of O. I. socorroensis,with which it bred sympatrically and temporally, in that a limited number of rock crevicesare available for nesting. The latter would nest in the small ones, and O. macrodactyla would take the larger. It is not clear why dark rumps would be selectedfor in lessmobile, coastalstorm- petrels (including O. markhami and O. tristrami in Peru and Hawaii, respectively). Of all the dark forms, only the Ashy Storm-Petrel has been studied intensively (Ainley et al. 1974), and an explanationof the colorphenomenon and all its correlates probably awaits studiesof other dark, coastalstorm-petrels. Crossin (1974) felt dark coloration made storm-petrels less visible to predators, especially the large gulls (Larus), which mostly occur near the coast. To me, white-rumped storm-petrelsare not more visible than dark ones, and I would think that predators would find this true as well. Furthermore, while frequenting offshore waters might reduce contact with gulls, it increasescontact with jaegers (particularly during winter), which are also potential predators. A relationshipto predation pressuresis thus not obvious. Selander (1971) reviewed several studiesthat related color to humidity, with darker- colored individuals occurring under more humid conditions. By remaining near the coastin California and Baja California (as does O. markhami in Peru, also), storm- petrels would encounter significantly more fog and overcast skies than those that frequent more offshore waters; this might also be true of O. tristrami in that it would maintain closer contact with the cloud forest conditions of the Hawaiian Islands. Along the Pacific coast of North America on the other hand, dark storm- petrels, O. leucorhoaor other species,nest only at siteswithin the Californian and Sonoran biogeographicalprovinces; in South America O. markhami nests in the very similar Pacific Desert province (seeUdvardy 1978; Fig. 6). The summer climate of these regions includes much fog and no rain; O. I. cheimomnestes, which lacks completely dark individuals, nests during the rainy season.Hence, there seemsto be a correlation between storm-petrel color and biotic provinces, but the relationship is not clear, becausebased on studiesof other birds (Selander1971), the darker birds should occur in the wetter conditions. Geographicvariation in O. leucorhoahas probably been affected by the amount 852 DAVIDG. AINLE¾ [Auk,Vol. 97

400'

60 50 40 $0

DEGREES NORTH LATITUDE Fig. 6. Mean distance (km) between breeding localities by latitude (data from Drent and Guiguet 1961, Sowls et al. 1979, Varoujean 1979) and the relationship of breeding latitudes to biogeographical provinces:shading, the Aleutian, Sitkan, and Oregonian rain forests; cross-hatching,the Californian coastalscrub; and white area, the Sonorandesert (seeUdvardy 1978).

of interbreeding among storm-petrelsfrom adjacent islands as a function of inter- island distance. Morphological charactersdiffered much more among birds from one nesting site to another in the southern region, where adjacent colonieswere separated by greater distancesthan in the northern region. Average distancesbetween colonies by latitude are graphed in Fig. 6, where it should be noted that north of British Columbia many additional but yet unknown nestingsites probably exist (Sowls et al. 1979). It is perhaps not surprising that most of the subspeciesof O. leucorhoa named at one time or another nest at single islands or island groups in the south, where one would expect lessenedinterchange among individuals from different colonies.

ACKNOWLEDGMENTS

I wish to thank the individuals and donors whose membershipsin and contributionsto the Point Reyes Bird Observatory ensurethe continued existenceof the Farallon researchstation. There this study began. I wish alsoto acknowledgethe help of personsand institutionswho made specimensand other information available to me: L. C. Binford (California Academy of Sciences),R. W. Campbell (Victoria Provincial Museum), N. K. Johnsonand V. M. Dziadosz (Museum of Vertebrate Zoology), S. L. Olson and R. L. Zusi (U.S. National Museum of Natural History), K. C. Parkes and J. Louglin (Carnegie Museum of Natural History), R. A. Paynter (Museum of Comparative Zoology), A.M. Rea (San Diego Museum of Natural History), R. W. Schreiber(Los AngelesCounty Museum), L. L. Short and J. Farrand (American Museum of Natural History), D. E. Willard (Field Museum of Natural History). S. M. Speich offered his unpublishedinformation. M. I. Cristy (Museum of Vertebrate Zoology)prepared the sonagramsand instructed me in their interpretation. S. L. Olson offered much encouragementand sorely neededadvice; G. E. Watson and W. L. Hoffman provided helpful comments on the MS. This is contribution 128 of the Point Reyes Bird Observatory. October1980] Leach's Storm-PetrelTaxonomy 853

LITERATURE CITED

AINLEY, D. G., S. MORRELL,& T. J. LEWIS. 1974. Patternsin the life historiesof storm petrelson the Farallon Islands. Living Bird 13: 295-312. , T. J. LEWIS, & S. MORRELL. 1976. Molt in Leach'sand Ashy storm-petrels.Wilson Bull. 88: 76-95. AMERICANORNITHOLOGISTS' UNION. 1957. Check-list of North American Birds, fifth ed. Baltimore, Amer. Ornithol. Union. 1973. Thirty-second supplement to the American Ornithologists'Union check-list of North American birds. Auk 90:411-419. AUSTIN, O. L., JR. 1952. Notes on some petrelsof the North Pacific. Bull. Mus. Comp. Zool. 107: 391-407. BOURNE,W. R. P. 1957. Additional noteson the birds of the Cape Verde Islands. Ibis 99: 182-190. , & J. WARHAM. 1966. Geographicalvariation in the Giant Petrelsof the genusMacronectes. Ardea 54: 45-67. BRIGGS,J. C. 1974. Marine zoogeography.New York, McGraw-Hill. CROSSIN,R. S. 1974. The storm petrels(Hydrobatidae). Pp. 154-203 in Pelagicstudies of seabirdsin the central and eastern Pacific Ocean (W. B. King, Ed.). SmithsonianContr. Zool. No. 158. DORST,J. 1962. The migration of birds. Boston, Houghton Mifflin Co. DRENT, R. H., & C. J. GUIGUET. 1961. A catalogueof British Columbia seabird colonies.British Columbia Provincial Mus., Occ. Pap. No. 12. HALL-CRAGGS,J., & P. J. SELLAR. 1976. Distinguishingcharacteristics in the burrow-callingof Storm and Leach's petrels. Brit. Birds 69: 293-297. HARRIS,M.P. 1969. The biologyof stormpetrels in the GalapagosIslands. Proc. California Acad. Sci., 4th Ser. 37: 95-166. LOOMIS,L. M. 1918. A review of the albatrosses,petrels, and diving petrels. Proc. California Acad. Sci., 4th Ser. 2: 1-187. MURPHY, R. C. 1951. The populationsof the Wedge-tailedShearwater (Puffinus pacificus). Amer. Mus. Novit. 1512: 1-21. NATIONAL MARINE FISHERIESSERVICE. 1971. Fishing information. May, June, and November Issues. La Jolla. PALMER,R. S. (Ed.). 1962. Handbookof North Americanbirds, vol. 1. New Haven, Connecticut,Yale Univ. Press. SELANDER,R. K. 1971. Systematicsand speciationin birds. Pp. 57-147 in Avian Biology, vol. 1. (D. S. Farner, J. R. King, and K. C. Parkes, Eds.). New York, AcademicPress. SMITH, W. J. 1977. The behaviorof communicating.Cambridge, Harvard Univ. Press. SOWLS,A. L., S. A. HATCH, & C. J. LENSINK. 1978. Catalogof Alaskanseabird colonies. U.S. Fish Wildl. Serv., FWS/OBS-78/78. STEEL,R. G. D., & J. H. TORRIE. 1960. Principlesand proceduresof statistics.New York, McGraw- Hill. SVERDRUP,H. U., M. W. JOHNSON,& R. H. FLEMING. 1942. The oceans:their physics,chemistry and general biology. Englewood Cliffs, New Jersey, Prentice-Hall. UDVARDY,M.D. F. 1978. World biogeographicalprovinces. Sausalito, California, CoEvolution Quart. VAN ROSSEM,A. J. 1942. Preliminary commenton somePacific coast petrels. Proc. Biol. Soc. Wash- ington 55: 9-12. VAROUJEAN,D. H. 1979. Seabirdcolony catalog: Washington, Oregon and California. Portland,Ore- gon, U.S. Fish Wildl. Serv., OBS.